The role of social projection in consumers' commonness fallacy

Recent research suggests that consumers often exhibit the “ commonness fallacy ” by overestimating the likelihood of others choosing a frequently consumed item (e.g., vanilla ice cream) rather than a less frequently consumed item (e.g., tiramisu). This research tests whether the extent to which consumers perceive their choices as similar to others — their social projection tendency — explains the commonness fallacy. Two preregistered studies ( N = 605) found that consumers with a higher social projection tendency overestimated the extent to which others would choose commonly consumed options if they themselves chose such options over less commonly consumed options. However, social projection cannot fully explain the commonness fallacy as participants with lower social projection tendencies also displayed the commonness fallacy in both studies. The findings delineate the commonness fallacy from consumers' well ‐ documented tendency to view others' choices as similar to their own.


| INTRODUCTION
Imagine that you are ordering ice cream for a party and are considering vanilla, a relatively commonly consumed ice cream flavor, and blueberry, a less commonly consumed flavor. You need to estimate how many guests would choose each flavor. Recent research suggests that people systematically exhibit the "commonness fallacy"-they overestimate others' interest in choosing a common option when it is pitted against a less common option (Reit & Critcher, 2020). If you are like the average participant in their studies, you will overestimate the proportion of guests who would prefer vanilla over blueberry ice cream.
Consumers often need to predict others' preferences, for instance, when buying gifts (Wooten, 2000). Such predictions are not limited to consumers; managers regularly need to predict customer preferences (Hoch, 1988;Weitz et al., 1986). However, it is hard to predict what others think. Consumer researchers have long documented people's tendency to mispredict others' preferences and choices (Barasz et al., 2016;David, 2018;Frederick, 2012;Gershoff et al., 2008;Givi & Das, 2021). A recurring finding in this literature is that people use their own behavior as a reference when predicting others' behavior (Krueger, 1998;Marks & Miller, 1987).
The commonness fallacy is distinct from such social projection.
Social projection or false-consensus bias implies that consumers think most others would like products they like. Such studies measure participants' preferences and predictions of others' preferences in isolation (say, for chocolate ice cream). However, the commonness fallacy involves specific choices between a frequently consumed (chocolate ice cream) and a less frequently consumed (blueberry) option. As Reit and Critcher (2020) documented, how common an option is may guide people's predictions in such situations. However, social projection may also play a role, and the commonness fallacy might merely reflect social projection-consumers predict that others will choose what they choose. In the present research, we aim to dissociate the two effects. Specifically, we test if commonness fallacy manifests in consumers with low social projection tendencies. If it does, it will give both researchers and practitioners greater confidence in the commonness fallacy as a heuristic independent of social projection. Reit and Critcher's (2020) studies found that people overestimate others' interest in choosing a commonly consumed option rather than a less commonly consumed one. In a typical study, they asked participants to choose between a common item and a less common item-helping them establish the actual choice shares. Additionally, they asked participants to forecast the percentage of other participants in the study who would choose each of the two options-a paradigm commonly used in consumer research (e.g., Frederick, 2012;Kurt & Inman, 2013). Across studies, participants systematically overestimated the choice share of the common item, thus exhibiting the commonness fallacy.

| Commonness fallacy
Why does this overestimation happen? It is probably true that frequently consumed items, such as vanilla ice cream, are more often chosen in the marketplace. An alternative possibility is that common items are cheaper to manufacture, which increases the supply of such items in the marketplace. Consumers buy common items more often because they are cheaper. Reit and Critcher (2020) argue that people erroneously use the perceived commonness of the item as a heuristic for predicting others' choices when the common item is presented along with a less common item. For instance, while vanilla flavor is more frequently consumed, imagine a consumer making a specific choice between vanilla and blueberry flavors. There is no reason to believe that just because most consumers choose vanilla from a relatively large assortment of ice creams available in a supermarket (maybe because it is cheaper), this particular consumer would also choose vanilla over blueberry in this specific instance when cost, convenience, and all other factors are held constant.

| Commonness fallacy as an egocentric bias
A rich body of consumer research literature has documented asymmetries between people's predictions of their own versus others' behaviors. For instance, consumers are more likely to overestimate how much others are willing to pay for a product than how much they would themselves pay (Frederick, 2012;Kurt & Inman, 2013). Similarly, while consumers like dissimilar products (e.g., horror movies and documentaries), they predict that others will dislike dissimilar products (Barasz et al., 2016). Further, consumers systematically believe that certain types of advertising will affect others more than themselves (Xie & Johnson, 2015;Yoon et al., 2000).
As consumers find it challenging to predict others' preferences accurately, they use a variety of strategies to ease this difficulty. They use others' previous choices to predict their future preferences (Barasz et al., 2016;Orhun & Urminsky, 2013). They also use attributes of the options or the decisions to make the predictions. For instance, when choosing for others, consumers tend to focus on how indulgent the product is (Laran, 2010). Similarly, as Reit and Critcher (2020) documented, consumers focus on how common each option is when predicting others' choices. As we discussed above, a third possibility is that people simply use their own preferences as anchors.
They might estimate that a greater percentage of others would like a product if they like the product (David, 2018;Gershoff et al., 2008). Reit and Critcher's (2020) studies overestimated the share of common options when they chose the common option; so social projection may explain commonness fallacy. However, imagine that in the opening vignette, the overall objective choice share of the vanilla flavor is 25%, and of blueberry, 75%. A consumer who has chosen blueberry has predicted that 40% of others would choose the vanilla flavor. While they did exhibit social projection (i.e., they predicted most others would choose blueberry), they still overestimated the choice share of vanilla flavor. Their predictions can be a function of social projection and some heuristic usage (e.g., the option's commonness). To delineate the effect of the two factors on commonness fallacy, we draw from the literature on social projection theory and the false consensus effect (Robbins & Krueger, 2005).

| Commonness fallacy and social projection
False consensus effect, or the belief that others' preferences are similar to one's own, can systematically affect consumer judgment.
Consumers might believe others would like the same products they like (Gershoff et al., 2008). When reading online reviews, they might believe that reviewers are similar to themselves (Naylor et al., 2011).
Such mistaken beliefs can affect their behavior. For instance, fearful investors sell off their stocks earlier if they believe that other investors share their risk attitude (Lee & Andrade, 2011). Research has shown that factors such as target knowledge (Lerouge & Warlop, 2006), interpersonal attachment styles (David, 2018), and empathy (Hattula et al., 2015) can affect social projection tendencies.
Drawing on this area of research, we argue that consumer differences in social projection tendencies might help us tease apart the effect of egocentric bias (i.e., false consensus effect) and options feature (i.e., perceived commonness of an option) in driving the commonness fallacy. We expect consumers with a high social projection tendency to project their own choices onto others. If they chose the commonly consumed (vs. less commonly consumed) option, they would predict that most others would also choose the commonly consumed (vs. less commonly consumed) option.
Consumers with low social projection tendencies may predict others' choices in two possible ways. First, such consumers might make accurate forecasts of others' choices as they are less likely to use their own choices as anchors. Past research suggests that sometimes, social forecasts can be quite accurate (Helzer & Dunning, 2012). For example, people's judgments about others' heights correlated with their own height, but the mean prediction was accurate (Ward, 1967).
If this is the case, we will find that commonness fallacy disappears when people don't rely on their own preferences, providing evidence for a wholly egocentric explanation for the phenomenon.
The other possibility is that instead of using their own choices as anchors, such consumers use some aspect of the choice contextperhaps the perceived commonness of the option-as a heuristic.
Many discrepancies between judgments of others' preferences versus ones' own can be explained over and above any effect predicted by mere social projection. For instance, consumers estimate how much others will pay for an item based on what they are willing to pay themselves, but also bias their estimates upwards, resulting in an overestimation (Frederick, 2012). Similarly, projection can explain only a part of the effect of indirect ("How much will another student donate?") versus direct ("How much would you donate?") survey items (Jang & Irwin, 2020). If commonness fallacy occurs over and above the social projection effect, we should observe commonness fallacy for consumers with low social projection tendencies.

| OVERVIEW OF STUDIES
We report two preregistered studies. In both studies, participants chose between pair(s) of commonly and less commonly chosen options and predicted others' choice(s). In Study 1, participants made a single choice between a common or less common candy. In Study 2, they made 11 choices spanning multiple product categories. Table 1 shows an overview of both studies. In both studies, we tested whether participants' choices and social projection tendencies interact to explain the commonness fallacy. We also tested if consumers with low social projection tendencies exhibit the commonness fallacy. We report all participants, conditions, and measures. Stimuli, data, and results are available at https://osf.io/ 74vdj/. See Supporting Information Materials for additional information.

| STUDY 1
In Study 1, we tested if the extent to which people overestimate the choice of common options based on their own choice depends on their level of social projection tendency. We preregistered the sample size, stimuli, measures, and analyses for this study on OSF: https:// osf.io/ht8v3.

| Predictions
We informed participants that we would ask 100 other participants in the study to choose between Milky Way Original and Milky Way Midnight Dark. We asked participants to estimate how many of these 100

| Social projection
Participants' social projection tendency is generally inferred from their choices and predictions in the literature. Therefore, there is a lack of a reliable individual difference measure for social projection.
As discussed above, both social projection and commonness fallacy can determine participants' predictions in the present context. Therefore, we created a two-item measure by following items measuring similarity between self and others developed by Toma et al. (2012). This is consistent with the key idea of social projection that people believe that most others would behave the same way as

| Role of social projection
To examine the role of social projection, we conducted a moderation analysis using Model 1 of the PROCESS macro for SPSS (Hayes, 2013). We

| Discussion
These results show that participants with a higher tendency to socially project overestimated the choice share of the common option when they chose the common (vs. less common) option. This difference attenuated for participants with a lower tendency to socially project.
However, participants with low social projection tendency did exhibit a statistically significant level of the commonness fallacy.

| STUDY 2
We had two aims in this study. Firstly, Study 1 used only one pair of products as stimuli. In this study, we aimed to increase the generalizability of the findings by using 11 pairs of frequently and less frequently consumed items. Second, the social projection measure in the previous study was context relevant. In this study, we used a general measure of social projection to examine if individual differences in social projection yield similar results as Study 1. We preregistered the sample size, stimuli, measures, and analyses for this study on OSF: https://osf.io/hpa9z.

| Procedure
We presented participants with 11 pairs of options. In each pair, one option was common, and the other was less common. The options were the same as those used in Reit and Critcher's (2020) Study 1.
We presented these 11 option pairs twice, once for predictions and once for choice. We randomized the presentation order of the choice task and the prediction task.

| Choices
We presented participants with 11 pairs of options, each on a separate page. We randomized the presentation order of the 11 pairs and the options within each pair. For each pair, we asked participants to choose one of the options. For instance, we asked participants, "If you had the choice of the following options for dinner tomorrow night, which would you choose?" Participants could indicate that they would choose either Thai food, the less common option, or Pizza, the common option.

| Predictions
We presented the same 11 pairs with the same randomization as described above. We told participants that we would present these pairs to 100 other participants in the study and that they must predict how many of those 100 participants would choose the two options. For instance, participants could indicate how many people would choose Thai food and Pizza such that the total added up to 100.

| Social projection
We adapted the items from Study 1 to measure the extent to which participants generally thought that others' choices would be similar to their own. We measured these items along with other demographic measures at the end of the study. Specifically, we asked participants  Table S1 in Supporting Information Materials for additional analysis as well as the prediction errors for all 11 pairs.

| Role of social projection
We conducted a moderation analysis using Model 1 of the PROCESS macro for SPSS (Hayes, 2013) and entered participants' average prediction error as the dependent variable (Y), the number of less common options they chose (mean-centered) as the independent variable (X), and their tendency to socially project (mean-centered) as the moderator (W). The overall model was significant   We also contribute to the research on consumers' misprediction of others' choices and preferences (Barasz et al., 2016;Frederick, 2012;Gershoff et al., 2008). Research in this area suggests that when predicting others' choices, consumers either anchor their predictions on their own preferences (e.g., Gershoff et al., 2008) or some attribute of an option such as how indulgent (Laran, 2010) or common (Reit & Critcher, 2020) the option is. We contribute to the growing body of research that both factors can have unique influences on consumers' predictions (e.g., Frederick, 2012;Jang & Irwin, 2020). Future researchers can examine if such joint effects can be found in other previously documented mispredictions, such as the misprediction of the effectiveness of advertising on others (e.g., Mo et al., 2018).
Given that commonness fallacy is related to social projection, our findings suggest that how commonness fallacy is related to other egocentric fallacies and biases needs more examination. For instance, research has documented various fallacies originating from consumers' construal level (Adler & Sarstedt, 2021). Other people are psychologically distant from consumers, which might affect the predictions consumers make about them. Similarly, another stream of research has documented how the presence of others can lead consumers to exhibit more egocentric biases, such as the spotlight effect (Roy et al., 2021). The extent to which commonness fallacy is related to these egocentric biases needs further investigation.
Our research also contributes to two important substantive domains for researchers and practitioners. The gift-giving literature has documented that people systematically mispredict the preferences of gift recipients (Givi & Das, 2021;Givi, 2020). Finally, another related substantive area is managers' predictions of consumers' choices. Research in this area suggests that managers often anchor on their own preferences when predicting consumer choice (Hattula et al., 2015;Herzog et al., 2021). Imagine a manager designing a product assortment or predicting the inventory requirements for the next sales cycle. Our findings suggest that even if managers don't rely on their own preferences in such decisions, they would still overpredict consumers' preferences for the frequently consumed items. Therefore, when designing interventions (e.g., Herzog et al., 2021), it is not sufficient to address egocentric biases.
To increase prediction accuracy, such interventions should jointly address multiple sources of prediction errors over and above egocentric sources. Future research can examine the effective design of such interventions.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on Open